Foodscapes of southern Ontario: neighbourhood deprivation and access to healthy and unhealthy food retail.
Polsky, Jane Y. ; Moineddin, Rahim ; Glazier, Richard H. 等
Unhealthy eating patterns and diet-related disease, such as obesity
and diabetes, are among the most pressing public health concerns in
Canada. Recent reports reveal that less than 1% of Canadians consume a
diet consistent with Canada's Food Guide recommendations, (1) and 6
in 10 are either overweight or obese. (2) Major changes in food
production and distribution in recent decades have flooded the consumer
food environment with highly palatable, processed, energy-dense and
low-cost foods, and are recognized to be the core environmental drivers
of increased energy intakes and ensuing rates of obesity. (3)
There is growing evidence that the neighbourhood retail
"foodscape" or food environment may play a role in shaping
dietary behaviours and diet-related health outcomes. (4-6) As well, the
availability of local food retail (FR) establishments has been shown to
vary according to level of neighbourhood socio-economic profile in urban
settings, with fewer retail sources of healthy foods (e.g.,
supermarkets) and more sources of unhealthy food (e.g., fast-food
outlets and conveniences stores) located in communities with higher
proportions of low-income and ethnic-minority residents compared with
more affluent neighbourhoods or those with fewer minorities. (4) This
pattern may, in part, account for the well-documented gradients in diet
and obesity by socio-economic status (SES). However, while such
disparities have been consistently documented in the United States,
systematic differences in FR distribution according to neighbourhood
socio-economic profile are less consistent and vary across settings
within other developed nations, including Canada. (4, 7-14)
Reasons for SES-related disparities in FR distribution are not well
specified in the health literature and have been rarely studied
empirically but likely include historical market forces, such as
pervasive supermarket restructuring resulting in closure of smaller
grocery stores in inner-city neighbourhoods, and location of fewer but
larger supermarkets in suburban areas offering larger lots and lower
rents. (15) The higher prevalence of fast food in lower-income,
inner-city communities may be due to purposeful targeting by food
retailers, lower rents and retail competition, and less restrictive
zoning practices. (16) Three recent Canadian investigations showed that
features of urban form, including higher population density and more
restrictive zoning (which separates residential areas from commercial
activity) were important predictors of FR distribution, beyond area
socio-economic characteristics. (8, 10, 17) Thus, there is a need for
additional Canadian research into disparities in FR distribution that
investigates other factors, such as characteristics related to
urbanization and zoning, that may help explain why FR is unevenly
distributed by area SES and thus help identify potential points of
intervention for public policy concerned with fostering healthier and
more equitable neighbourhoods. Additionally, there is a need for more
comprehensive FR assessment that moves beyond the common approach of
measuring access to just one or two types of retail establishment (8, 9,
13, 17) in order to achieve a more complete representation of the
complex set of FR options available to residents. (14)
Accordingly, our study aims to assess whether access to retail
sources of healthy and unhealthy food varies systematically according to
level of neighbourhood deprivation in three diverse settings in southern
Ontario, a region that has, to date, received little attention in the
empirical literature. Specifically, this study uses several
comprehensive measures of the local foodscape to assess whether access
to healthy and unhealthy FR varies according to level of neighbourhood
material deprivation, and whether urban form characteristics explain
these associations.
METHODS
Study design and area
For this cross-sectional, ecological study we focused on three
diverse urban settings in southern Ontario: the City of Toronto,
Canada's largest metropolitan centre; the City of Brampton and the
City of Mississauga, two contiguous and largely suburban cities adjacent
to the west of Toronto; and the City of Hamilton, a medium-size city
located southwest of Brampton/Mississauga. Because of data limitations,
only neighbourhoods located within the old municipal boundaries of
Hamilton were included. Urban census tracts (CTs) have been shown to be
good proxies for naturally defined neighbourhoods in previous
health-related research, (18) and 2006 CTs were used as neighbourhood
units of analysis in this study.
Measures of FR access
All FR records and their geographic coordinates were obtained from
a single proprietary commercial database (Dun & Bradstreet Canada,
Inc.), which contained a comprehensive inventory of all FR outlets
located within the study area and a 5-km buffer of each region's
boundaries in January 2008. A list of food stores and restaurants was
initially extracted using the North American Industry Classification
codes, followed by extensive cleaning to remove duplicate listings and
defunct businesses, as well as additional reclassification efforts in a
protocol highly consistent with other researchers. (19) We then
classified supermarkets, grocery stores and fruit and vegetable shops as
"healthy" and fast-food outlets and convenience stores as
"unhealthy" FR. This classification scheme is similar to
previous research (5, 6, 20, 21) and is summarized in Table 1, along
with definitions of each FR type.
We calculated three measures of access to healthy and unhealthy FR
outlets within a 10-minute walk of where most people lived within
neighbourhoods using network analysis tools in ArcGIS 9.3 software
(ESRI, Redlands, CA). Our assessment of "access" incorporated
the dimensions of availability (presence and number of outlets) and
accessibility (proximity to outlets), (22) and proceeded as follows:
first, in order to better account for population distribution within
each neighbourhood, (23) we calculated the number of healthy and
unhealthy outlets accessible by street network within a 720-metre buffer
(a 10-minute walking distance at an estimated speed of 1.2 m/sec) around
the centroid of each residential census dissemination block contained
within a neighborhood's boundaries (region average of 20 blocks per
neighbourhood). Similar to previous studies, (8, 11, 12) we then
calculated median numbers of healthy and unhealthy outlets within each
neighbourhood, weighted by block population, to serve as two summary
measures of absolute access to 1) healthy and 2) unhealthy FR. Our third
measure assessed relative access to unhealthy FR and was calculated as
the percentage of all FR outlets that were unhealthy. (24)
Measure of neighbourhood material deprivation
Neighbourhood deprivation was measured using the material
deprivation dimension of the 2006 Ontario Marginalization Index, a
theoretically informed, empirically derived and validated composite
index of Canadian marginalization constructed using measures from the
2006 Canadian Census. (25) The six component measures of this material
deprivation index are summarized in Table 2. For ease of interpretation,
we created population-weighted region-specific quintiles of
neighbourhood material deprivation, which were collapsed into a
three-level variable for regression models because of sparse data: most
deprived (quintiles 1, 2), middle (quintile 3) and least deprived
neighbourhoods (quintiles 4, 5).
Measures of urban form
Using neighbourhood-level data from the 2006 Canadian Census, we
calculated population density per square kilometre of land area,
percentage of employed population aged 15+ who used public transit to
commute to work and percentage of occupied private dwellings that were
single detached houses. These indicators were conceptualized as proxy
measures of zoning and urbanization, and were significant predictors of
neighbourhood access to FR establishments in recent Canadian studies.
(8, 17)
Statistical analyses
Regression models unadjusted and adjusted for urban form factors
assessed the association of material deprivation with each measure of FR
access separately. As a result of evidence of interaction between region
and deprivation for some measures of FR, all analyses were stratified by
region. Negative binomial regression was used to model the number of
healthy and unhealthy FR outlets (i.e., absolute access to healthy and
unhealthy FR). The results of these models are expressed as rate ratios
(RRs), along with their associated 95% confidence intervals (CIs). Tobit
regression for censored data (26) was used to model the percentage of
unhealthy outlets (i.e., relative access to unhealthy FR), yielding
slope coefficients that can be interpreted in a manner similar to linear
regression. These models specified a lower bound of zero to include as
censored observations neighbourhoods with a zero denominator (i.e.,
censored for percentage of unhealthy outlets because of lack of any
healthy or unhealthy FR for 8 neighbourhoods in Toronto, 12 in
Brampton/Mississauga and 4 in Hamilton). A number of sensitivity
analyses were also conducted, using alternative modeling approaches and
definitions of FR access (i.e., alternative definitions of unhealthy and
healthy FR outlets; presence of any outlets; density of outlets per 10,
000 residents within census tract boundaries; and access within 3-km
buffers). These revealed results highly consistent with those of the
final analyses. All analyses were conducted in SAS Version 9.3 (SAS,
Cary, NC).
Ethics approval was obtained from the University of Toronto and St.
Michael's Hospital Offices of Research Ethics.
RESULTS
Table 2 describes demographic, socio-economic and urban form
characteristics of 524 study neighbourhoods in Toronto (population of 2,
501, 502), 185 in Brampton/Mississauga (population of 1, 023, 450) and
95 in Hamilton (population of 327, 278). In all study regions,
neighbourhoods classified as more deprived according to the material
deprivation index had consistently higher levels of socio-economic
disadvantage across its component measures, including a higher
proportion of lone-parent families and homes needing major repairs, as
well as significantly lower household income (region average of $40, 867
vs. $92, 625 in the most vs. least deprived areas).
In all study regions, unhealthy FR was more readily accessible in
absolute terms (i.e., number of unhealthy outlets within a 10-minute
walk) than healthy FR, regardless of level of material deprivation
(results not shown). Absolute access to both healthy and unhealthy FR
differed by level of material deprivation in every region (all p values
<0.01), with more outlets found in areas with higher levels of
deprivation (median of 0 vs. 1 healthy outlets and 0 vs. 3 unhealthy
outlets in the least vs. most deprived areas, overall). Relative access
to unhealthy FR, expressed as percentage of outlets that were unhealthy,
was high across the study regions, ranging from 71.1% to 85.7%, but
showed no significant differences by level of neighbourhood deprivation.
In unadjusted analyses for all study regions, the most deprived
neighbourhoods had a 2- to 4-fold greater median number of healthy FR
outlets than the least deprived areas (Table 3). After adjustment for
urban form factors, these associations were substantially attenuated and
remained significant only for Toronto.
Patterns for absolute access to unhealthy FR in relation to
material deprivation differed across study regions (Table 4): in
unadjusted analysis for Toronto, higher levels of deprivation were
related to fewer unhealthy food retailers (RR=0.90; 95% CI: 0.69-1.18
for most vs. least deprived areas) but substantially more unhealthy
outlets in Brampton/Mississauga (RR=2.29; 95% CI: 1.21-4.31) and
Hamilton (RR=4.45; 95% CI: 2.78-7.13). For the latter two regions,
inclusion of urban form measures substantially attenuated the effect of
material deprivation, reducing the RR to null in Brampton/ Mississauga;
however, deprivation remained significantly related to increased
absolute access to unhealthy FR in Hamilton's middle (RR=1.61; 95%
CI: 1.00-2.59) and most deprived (RR=2.11; 95% CI: 1.36-3.28) vs. least
deprived areas. In Toronto, the adjusted effect of deprivation remained
negative for both middle and highest levels of material deprivation
(RR=0.7, p<0.05).
In all unadjusted analyses, urban form measures were significantly
related to absolute access to both healthy and unhealthy FR (all p
values<0.001; Tables 3 and 4). In all regions, population density and
public transit use were positively related to absolute FR access, while
proportion of detached housing had a negative association. In adjusted
models, estimates for population density and public transit use
diminished consistently for all study regions, whereas estimates for
proportion of detached housing remained relatively unchanged.
Relative access to less healthy FR (percentage of outlets that were
unhealthy) had no clear relationship with either neighbourhood
deprivation or urban form characteristics (Table 5). For material
deprivation, the only significant association was seen in Toronto,
where, on average, the most deprived areas had a 4.3% lower proportion
of unhealthy FR outlets than the least deprived areas, after accounting
for urban form factors.
DISCUSSION
This study adds to the growing Canadian FR environment literature
by describing the distribution of commonly accessed FR establishments
across more and less materially deprived neighbourhoods in three
geographically and socially diverse regions in southern Ontario. Across
all study regions, the most materially deprived neighbourhoods had 2 to
4 times more stores selling healthy (supermarkets, grocery stores, and
fruit and vegetable stores) and unhealthy food (fast-food and
convenience stores) within a 10-minute walk of where most people lived
compared with the least deprived areas, except in Toronto, where
unhealthy FR was more plentiful in less deprived areas. Urban form
factors, such as population density and proportion of detached
dwellings, helped to explain these associations for Brampton/
Mississauga and Hamilton more so than for Toronto.
In addition to measures of absolute access (number of outlets),
this study is among the first to assess relative access to unhealthy FR
(unhealthy food outlets as a percentage of both healthy and unhealthy
FR) according to level of neighbourhood deprivation. We found that
relative access to unhealthy FR was generally unrelated to level of
neighbourhood material deprivation in Brampton/Mississauga and Hamilton.
In Toronto, the average proportion of unhealthy FR was 4.3% lower in the
most vs. least deprived areas.
Our findings are consistent with those of recent reports from
Montreal, Edmonton and urban areas of British Columbia in demonstrating
that areas with a higher proportion of materially deprived residents are
not systematically disadvantaged in terms of absolute access to retail
sources of fresh produce, such as supermarkets, grocery stores, and
fruit/vegetable shops, and have similar or better access compared to
more affluent areas. (8-11, 14, 17) In contrast, a recent study from
London, Ontario, showed worse access to supermarkets among residents of
lower SES neighbourhoods, which supports a need for future
region-specific examinations of FR distribution and consideration of its
unique demographic, economic and political histories and profiles. (12)
Our findings also resonate with recent investigations in Montreal
and Edmonton in showing higher levels of exposure in more
socio-economically disadvantaged areas to retail sources of unhealthy
food (fast-food outlets and conveniences stores)-areas also known as
"food swamps." (7, 8, 13, 14) In terms of relative access to
unhealthy FR, our results were generally consistent with the only
analogous study in the literature, to our knowledge, which found that
residents of the poorest areas in Montreal were exposed to fewer
fast-food restaurants as a proportion of all restaurants in the area,
and more fruit and vegetable stores as a proportion of all stores,
relative to the wealthiest areas. (14) Collectively, these findings show
little support for the notion raised by some US researchers that
socio-economically disadvantaged neighbourhooods may be systematically
targeted by unhealthy food retailers, (16) and indicate that such areas
generally provide better local access to FR establishments of all types,
given their commonly higher levels of density and mixed land use.
With regard to the last point, we showed that higher population
density and lower proportion of single detached dwellings (proxy
measures of areas zoned for mixed residential and commercial use) and
higher levels of public transit use (a possible attraction for FR site
selection (17)) were indeed related to better FR access in absolute
terms, and adjustment for these factors explained much of the
association between neighbourhood deprivation and number of food outlets
within walking distance in Brampton/Mississauga and to a lesser extent
Hamilton. These findings are in line with existing research from British
Columbia, Quebec and Alberta, (8, 10, 17) and suggest that urban
planning factors likely play a role in shaping the distribution of FR in
urban areas, perhaps more so than area socio-demographic composition. In
other words, socio-economically disadvantaged residents may be more
likely to find affordable housing in more densely populated areas with
higher levels of retail density, which commonly includes diverse types
of FR. In contrast, wealthier residents have the option of attaining
some level of "isolation" from both more and less desirable
types of FR by settling in more sparsely populated areas zoned
exclusively for residential purposes. (8) Interestingly, the same urban
form correlates had little impact on the neighbourhood deprivation-FR
access association in Toronto. This may suggest that urban form factors
other than those included in our study relate more strongly to siting
and distribution of FR, or reflect considerable variation in urban form
characteristics between neighbourhoods of similar SES profile. For
example, while many of Toronto's low-income residents live in
neighbourhoods with high levels of food and other retail activity, a
large proportion resides in clusters of post-war high-rise apartment
buildings surrounded by open space and few retail amenities. (27)
Several limitations of this research deserve mention. As in other
ecological analyses, our discussion is limited to the association
between deprivation and FR access at the level of the neighbourhood and
not among individuals, and we were unable to assess the impact of such
area-based measures on individuals' food purchasing or dietary
behaviours. Additionally, the study assessed only two dimensions of FR
access (availability and accessibility). Other dimensions include
financial affordability and socio-cultural acceptability of foods to
consumers, which are also important in shaping food choices. (22)
Further, our use of store type as a proxy measure for access to
"healthy" and "unhealthy" food is a limitation that
may have introduced some degree of misclassification into our assessment
of FR exposure. In reality, stores and restaurants offer a variety of
more and less healthy foods. For example, supermarkets stock many
unhealthy, processed food products; however, they are also the most
important and reliable source of affordable fresh produce for consumers.
(28) Similarly, while many fast-food restaurants offer healthier,
"better-for-you" menu choices, traditional energy-rich,
nutritionally deficient foods remain the dominant default within most
fast-food establishments and are the foods most heavily marketed both
outside and inside restaurants. (29) Regular consumption of fast food
has been linked to weight gain and other adverse cardio-metabolic
outcomes. (30, 31)
This study's finding that more socio-economically
disadvantaged areas did not have systematically better relative access
to unhealthy food is a positive one from a social equity perspective.
However, our finding of plentiful access to unhealthy food sources in
absolute terms (alongside better absolute access to healthy FR) in
Hamilton and Brampton/Mississauga is a concern, given recent research
linking a greater share of unhealthy FR to poorer dietary patterns and
higher levels of obesity and diabetes. (5, 6, 21) There are also growing
concerns that "food swamps" may influence food choices more
strongly than poor access to sources of healthy food. (7) Additionally,
residents of lower-income areas may be more vulnerable to their
surrounding FR environment because of limited transportation options,
time and financial constraints, and value for money of fast food. (13,
32)
Evidence of plentiful access to sources of unhealthy food
(particularly in low-income areas) from this and other studies indicates
that better eating and obesity prevention strategies in Canadian urban
areas could concentrate on encouraging healthier menu offerings within
the highly accessible convenience stores and fast-food outlets to ensure
that there is a mix of FR outlets maximizing exposure to healthy food
choices. Future research using more comprehensive FR measures is needed
to better understand how various population groups interact with their
local foodscape, and which policies will have a positive impact on food
choices and diet-related health outcomes.
Acknowledgements: All analyses were conducted at the Centre for
Research on Inner City Health, Li Ka Shing Knowledge Institute of St.
Michael's Hospital. During the tenure of this study, Jane Polsky
was supported by a Doctoral Research Award from the Canadian Institutes
of Health Research. Richard Glazier is funded as a Clinician Scientist
in the Department of Family and Community Medicine at the University of
Toronto and St. Michael's Hospital. Gillian Booth holds Mid-Career
Investigator Awards from the Heart and Stroke Foundation of Ontario and
the Department of Medicine at the University of Toronto. The authors
thank Jonathan T. Weyman for creating all GIS-based measures for this
work. All views expressed in this paper are those of the authors and do
not necessarily reflect the views of any affiliated organization.
Conflict of Interest: None to declare.
REFERENCES
(1.) Garriguet D. Diet quality in Canada. Health Reports 2009;
20(3): 1-12.
(2.) Obesity in Canada: A Joint Report from the Public Health
Agency of Canada and the Canadian Institute for Health Information.
Ottawa, ON, 2011.
(3.) Swinburn BA, Sacks G, Hall KD, McPherson K, Finegood DT,
Moodie ML, et al. The global obesity pandemic: Shaped by global drivers
and local environments. Lancet 2011; 378(9793): 804-14.
(4.) Black C, Moon G, Baird J. Dietary inequalities: What is the
evidence for the effect of the neighbourhood food environment? Health
Place 2014; 27: 229-42.
(5.) Mercille G, Richard L, Gauvin L, Kestens Y, Shatenstein B,
Daniel M, et al. Associations between residential food environment and
dietary patterns in urban-dwelling older adults: Results from the
VoisiNuAge study. Public Health Nutr 2012; 15(11): 2026-39.
(6.) Spence JC, Cutumisu N, Edwards J, Raine KD, Smoyer-Tomic K.
Relation between local food environments and obesity among adults. BMC
Public Health 2009; 9: 192.
(7.) Measuring the Food Environment in Canada. Ottawa, ON: Health
Canada, 2013. Report No. 120182.
(8.) Smoyer-Tomic KE, Spence JC, Raine KD, Amrhein C, Cameron N,
Yasenovskiy V, et al. The association between neighborhood socioeconomic
status and exposure to supermarkets and fast food outlets. Health Place
2008; 14(4): 740-54.
(9.) Smoyer-Tomic KE, Spence JC, Amrhein C. Food deserts in the
Prairies? Supermarket accessibility and neighborhood need in Edmonton,
Canada. Professional Geographer 2006; 58(3): 307-26.
(10.) Daniel M, Kestens Y, Paquet C. Demographic and urban form
correlates of healthful and unhealthful food availability in Montreal,
Canada. Can J Public Health 2009; 100(3): 189-93.
(11.) Apparicio P, Cloutier M, Shearmur R. The case of
Montreal's missing food deserts: Evaluation of accessibility to
food supermarkets. Int J Health Geogr 2007; 6(1): 4.
(12.) Larsen K, Gilliland J. Mapping the evolution of 'food
deserts' in a Canadian city: Supermarket accessibility in London,
Ontario, 1961-2005. Int J Health Geogr 2008; 7.
(13.) Hemphill E, Raine K, Spence JC, Smoyer-Tomic KE. Exploring
obesogenic food environments in Edmonton, Canada: The association
between socioeconomic factors and fast-food outlet access. Am J Health
Promot 2008; 22(6): 426-32.
(14.) Mercille G, Richard L, Gauvin L, Kestens Y, Payette H, Daniel
M. Comparison of two indices of availability of fruits/vegetable and
fast food outlets. J Urban Health 2013; 90(2): 240-45.
(15.) Eisenhauer E. In poor health: Supermarket redlining and urban
nutrition. GeoJournal 2001; 53(2): 125-33.
(16.) Kwate NOA. Fried chicken and fresh apples: Racial segregation
as a fundamental cause of fast food density in black neighborhoods.
Health Place 2008; 14(1): 32-44.
(17.) Black JL, Carpiano RM, Fleming S, Lauster N. Exploring the
distribution of food stores in British Columbia: Associations with
neighbourhood socio-demographic factors and urban form. Health Place
2011; 17(4): 961-70.
(18.) Ross NA, Tremblay S, Graham K. Neighbourhood influences on
health in Montreal, Canada. Soc Sci Med 2004; 59(7): 1485-94.
(19.) Ohri-Vachaspati P, Martinez D, Yedidia MJ, Petlick N.
Improving data accuracy of commercial food outlet databases. Am J Health
Promot 2011; 26(2): 116-22.
(20.) Mason KE, Bentley RJ, Kavanagh AM. Fruit and vegetable
purchasing and the relative density of healthy and unhealthy food
stores: Evidence from an Australian multilevel study. J Epidemiol
Community Health 2013; 67(3): 231-36.
(21.) Designed for disease: The link between local food
environments and obesity and diabetes. Davis, CA: Regents of University
of California, PolicyLink and California Center for Public Health
Advocacy, 2008.
(22.) Caspi CE, Sorensen G, Subramanian SV, Kawachi I. The local
food environment and diet: A systematic review. Health Place 2012;
18(5): 1172-87.
(23.) Apparicio P, Abdelmajid M, Riva M, Shearmur R. Comparing
alternative approaches to measuring the geographical accessibility of
urban health services: Distance types and aggregation-error issues. Int
J Health Geogr 2008; 7: 7.
(24.) Truong K, Fernandes M, An R, Shier V, Sturm R. Measuring the
physical food environment and its relationship with obesity: Evidence
from California. Public Health 2010; 124(2): 115-18.
(25.) Matheson FI, Dunn JR, Smith KLW, Moineddin R, Glazier RH.
Development of the Canadian Marginalization Index: A new tool for the
study of inequality. Can J Public Health 2012; 103(Suppl 2): S12-S16.
(26.) Long JS. Regression Models for Categorical and Limited
Dependent Variables. Thousand Oaks, CA: Sage Publications, 1997.
(27.) Toward Healthier Apartment Neighbourhoods: A Healthy Toronto
by Design Report. Toronto, ON: Centre for Urban Growth and Renewal &
Toronto Public Health, September 2012.
(28.) Tarasuk V. Policy directions to promote healthy dietary
patterns in Canada. Appl Physiol Nutr Metab 2010; 35(2): 229-33.
(29.) Harris JL, Schwartz MB, Brownell KD. Fast Food FACTS:
Evaluating Fast Food Nutrition and Marketing to Youth. Yale Rudd Center
for Food Policy and Obesity, 2010.
(30.) Duffey KJ, Gordon-Larsen P, Steffen LM, Jacobs Jr DR, Popkin
BM. Regular consumption from fast food establishments relative to other
restaurants is differentially associated with metabolic outcomes in
young adults. J Nutr 2009; 139(11): 2113-18.
(31.) Pereira MA, Kartashov AI, Ebbeling CB, Van Horn L, Slattery
ML, Jacobs Jr. PDR, et al. Fast-food habits, weight gain, and insulin
resistance (the CARDIA study): 15-year prospective analysis. Lancet
2005; 365(9453): 36-42.
(32.) Boone-Heinonen J, Gordon-Larsen P, Kiefe CI, Shikany JM,
Lewis CE, Popkin BM. Fast food restaurants and food stores-longitudinal
associations with diet in young to middle-aged adults: The CARDIA study.
Arch Intern Med 2011; 171(13): 1162-70.
Received: April 21, 2014
Accepted: July 25, 2014
Jane Y. Polsky, MSc, [1, 2] Rahim Moineddin, PhD, [3, 5] Richard H.
Glazier, MD, [1-5] James R. Dunn, PhD, [2, 5, 6] Gillian L. Booth, MD
[2, 5, 7]
Author Affiliations
[1.] Dalla Lana School of Public Health, University of Toronto,
Toronto, ON
[2.] Centre for Research on Inner City Health, Li Ka Shing
Knowledge Institute of St. Michael's Hospital, Toronto, ON
[3.] Department of Family and Community Medicine, University of
Toronto, Toronto, ON
[4.] Department of Family and Community Medicine, St.
Michael's Hospital, Toronto, ON
[5.] Institute for Clinical Evaluative Sciences, Toronto, ON
[6.] Department of Health, Aging and Society, McMaster University,
Hamilton, ON
[7.] Department of Medicine, University of Toronto, Toronto, ON
Correspondence: Jane Y. Polsky, Centre for Research on Inner City
Health, St. Michael's Hospital, 30 Bond St, Toronto, ON M5B 1W8,
Tel: 416-864-6060, ext. 77486, E-mail: jane.polsky@mail.utoronto.ca
Table 1. Study definitions of food retail types
Food retail type Definition
Healthy food retailer
Supermarket Large national/regional stores offering a
full line of grocery products (food and
non-food). For independently owned stores,
only stores with annual sales volumes of $2
million or more were included.
Grocery store All food stores, other than supermarkets
and convenience/variety stores or specialty
food stores, offering a line of dry
grocery, canned goods and/or perishable
food items.
Fruit & vegetable shop Smaller stores specializing in the sale of
fresh fruits and vegetables. Unhealthy food
retailer
Convenience or Smaller stores offering a limited variety
variety store of grocery and daily living items (food and
non-food) open outside normal business
hours and days. Stores within gas stations
were included.
Fast-food outlet Locally owned or franchised limited-service
restaurants (i.e., venues without table
service where patrons pay before receiving
their meal). Outlets that do not serve full
meals (e.g., coffee or snack shops, cafes)
were excluded.
Table 2. Socio-economic and urban form characteristics of
neighbourhoods by quintile of material deprivation from 2006
Canadian Census (mean values)
Toronto
Q1 Q2 Q3 Q4 Q5
Neighbourhoods, 116 102 99 103 104
N
Socio-economic
characteristics *
% Aged 25-64 without 3.8 8.3 12.1 17.8 21.9
certificate, diploma
or degree
% Lone-parent 12.5 16.5 19.6 22.3 29.9
families
% Unemployed 5.3 6.8 7.5 8.2 10.5
% Homes needing 4.9 6.3 7.7 8.9 11.5
major repairs
% Low-income 12.8 19.5 23.6 26.7 36.4
population
% Population 5.1 9.3 11.9 14.5 18.9
receiving gov't
transfer payments
Median household 88,063 63,068 54,842 48,523 39,786
income ($)
Urban form
characteristics
Population 5802 6200 6553 7022 8363
density (per
[km.sup.2])
% Detached housing 41.5 38.0 33.0 23.2 17.8
% Population 29.2 31.6 33.4 36.7 39.1
using public
transit
Brampton/Mississauga
Q1 Q2 Q3 Q4 Q5
Neighbourhoods, 30 36 38 39 42
N
Socio-economic
characteristics *
% Aged 25-64 without 7.1 10.5 10.9 14.8 16.3
certificate, diploma
or degree
% Lone-parent 10.7 13.1 16.1 17.8 21.1
families
% Unemployed 5.3 6.0 6.1 7.0 8.2
% Homes needing 1.7 2.6 3.4 4.7 7.9
major repairs
% Low-income 10.2 11.5 14.2 16.2 22.3
population
% Population 5.2 6.8 7.8 10.1 12.4
receiving gov't
transfer payments
Median household 97,211 85,832 75,351 67,415 54,384
income ($)
Urban form
characteristics
Population 3874 3558 4085 4301 5429
density (per
[km.sup.2])
% Detached housing 66.6 56.7 45.1 42.8 23.2
% Population 11.4 12.6 14.2 13.9 18.2
using public
transit
Hamilton
Q1 Q2 Q3 Q4 Q5
Neighbourhoods, 18 19 19 19 20
N
Socio-economic
characteristics *
% Aged 25-64 without 9.3 15.8 14.9 21.6 30.9
certificate, diploma
or degree
% Lone-parent 13.0 18.3 20.0 23.5 30.5
families
% Unemployed 5.3 5.9 7.6 8.5 10.6
% Homes needing 4.4 6.7 8.4 9.5 13.2
major repairs
% Low-income 9.7 16.8 20.4 28.4 38.0
population
% Population 9.8 13.5 15.3 17.6 22.8
receiving gov't
transfer payments
Median household 70.929 57,634 47,529 43,367 35,543
income ($)
Urban form
characteristics
Population 2439 3803 3851 4252 4745
density (per
[km.sup.2])
% Detached housing 77.1 59.1 54.9 46.3 45.5
% Population 6.4 10.1 13.7 14.3 18.5
using public
transit
Q1=least deprived and Q5=most deprived neighbourhood.
* All socio-economic indicators listed, except for median household
income (before tax), were components of the material deprivation
index used to create quintiles of neighbourhood deprivation. The
index components were: 1) % of persons aged 25-64 without
certificate, diploma or degree; 2) % of families who are
lone-parent families; 3) % of persons aged 15+ who are unemployed;
4) % of households living in homes in need of major repairs; 5) %
of the population considered low income (i.e., living below the
low-income cut-off, before tax, in 2005); and 6) % of the
population receiving government transfer payments.
Table 3. Negative binomial regression modeling absolute access to
healthy food retailers (i.e., number of healthy outlets) within a
10-minute walk of residential areas in neighbourhoods
Toronto
Adjusted
Unadjusted model ([double
models ([dagger]) dagger])
RR 95% CI RR 95% CI
Material
deprivation
Least deprived 1.00 1.00
Middle 1.18 (0.78, 1.78) 1.05 (0.71, 1.56)
Most deprived 2.07 *** (1.50, 2.84) 1.37 * (1.01, 1.87)
Urban form
Population density
(1000 per
[km.sup.2]) 1.16 *** (1.11, 1.20) 1.04 * (1.00, 1.07)
% Detached housing 0.97 *** (0.96, 0.98) 0.98 *** (0.97, 0.99)
% Public transit
users 1.06 *** (1.05, 1.08) 1.03 *** (1.01, 1.04)
Brampton/Mississauga
Unadjusted Adjusted
models ([dagger]) model ([dagger])
RR 95% CI RR 95% CI
Material
deprivation
Least deprived 1.00 1.00
Middle 1.30 (0.44, 3.85) 0.56 (0.17, 1.86)
Most deprived 1.97 (0.82, 4.69) 1.08 (0.41, 2.86)
Urban form
Population density
(1000 per
[km.sup.2]) 1.18 ** (1.05, 1.32) 1.09 (0.98, 1.23)
% Detached housing 0.97 *** (0.96, 0.99) 0.99 (0.97, 1.01)
% Public transit
users 1.11 *** (1.05, 1.17) 1.07 (1.00, 1.15)
Hamilton
Unadjusted Adjusted
models ([dagger]) model ([dagger])
RR 95% CI RR 95% CI
Material
deprivation
Least deprived 1.00 1.00
Middle 2.19 (0.75, 6.37) 1.07 (0.42, 2.70)
Most deprived 4.39 *** (1.86,10.36) 1.85 (0.84, 4.05)
Urban form
Population density
(1000 per
[km.sup.2]) 1.33 *** (1.19, 1.49) 1.08 (0.97, 1.19)
% Detached housing 0.96 *** (0.95, 0.97) 0.98 *** (0.96, 0.99)
% Public transit
users 1.14 *** (1.09, 1.20) 1.04 (0.99, 1.08)
RR=rate ratio; 95% CI=95% confidence interval.
([dagger]) Estimates obtained from separate unadjusted negative
binomial models.
([double dagger]) All estimates mutually adjusted for all other
variables listed.
* p<0.05, ** p<0.01, *** p<0.001.
Table 4. Negative binomial regression modeling absolute access to
unhealthy food retailers (i.e., number of unhealthy outlets) within a
10-minute walk of residential areas in neighbourhoods
Toronto
Adjusted
Unadjusted model ([double
models ([dagger]) dagger])
RR 95% CI RR 95% CI
Material
deprivation
Least deprived 1.00 1.00
Middle 0.63 ** (0.44, 0.88) 0.69 * (0.51, 0.92)
Most deprived 0.90 (0.69, 1.18) 0.73 * (0.57, 0.93)
Urban form
Population density
(1000 per
[km.sup.2]) 1.12 *** (1.09, 1.15) 1.04 ** (1.01, 1.06)
% Detached housing 0.97 *** (0.96, 0.97) 0.97 *** (0.97, 0.98)
% Public transit
users 1.04 *** (1.03, 1.06) 1.02 ** (1.01, 1.03)
Brampton/Mississauga
Unadjusted Adjusted
models ([double dagger]) model ([dagger])
RR 95% CI RR 95% CI
Material
deprivation
Least deprived 1.00 1.00
Middle 1.31 (0.59, 2.89) 0.71 (0.33, 1.55)
Most deprived 2.29 * (1.21, 4.31) 1.00 (0.52, 1.95)
Urban form
Population density
(1000 per
[km.sup.2]) 1.12 ** (1.03, 1.21) 1.04 (0.98, 1.11)
% Detached housing 0.97 *** (0.96, 0.98) 0.99 (0.97, 1.00)
% Public transit
users 1.13 *** (1.08, 1.18) 1.09 ** (1.03, 1.14)
Hamilton
Adjusted
Unadjusted model ([double
models ([dagger]) dagger])
RR 95% CI RR 95% CI
Material
deprivation
Least deprived 1.00 1.00
Middle 2.40 ** (1.34, 4.27) 1.61 * (1.00, 2.59)
Most deprived 4.45 *** (2.78, 7.13) 2.11 *** (1.36, 3.28)
Urban form
Population density
(1000 per
[km.sup.2]) 1.29 *** (1.17, 1.42) 1.12 ** (1.05, 1.19)
% Detached housing 0.98 *** (0.98, 0.99) 1.00 (1.00, 1.01)
% Public transit
users 1.14 *** (1.11, 1.17) 1.09 *** (1.05, 1.12)
RR=rate ratio; 95% CI=95% confidence interval.
([dagger]) Estimates obtained from separate unadjusted negative
binomial models.
([double dagger]) All estimates mutually adjusted for all other
variables listed.
* p<0.05, ** p<0.01, *** p<0.001.
Table 5. Tobit regression modeling relative access to unhealthy food
retailers (i.e., % of outlets that were unhealthy) within a 10-minute
walk of residential areas in neighbourhoods
Toronto
Adjusted
Unadjusted model ([double
models ([dagger]) dagger)]
[beta] 95% CI [beta] 95% CI
Material
deprivation
Least deprived -- --
Middle -0.34 (-4.23, 3.54) 0.93 (-4.30, 2.44)
Most deprived -2.27 (-5.39, 0.85) -4.28 * (-8.15, -0.42)
Urban form
Population
density
(1000 per
[km.sup.2]) 0.31 * (0.07, 0.55) 0.15 (-0.16, 0.46)
% Detached
housing -0.08 ** (-0.14, -0.03) 0.09 * (-0.16, -0.02)
% Public
transit 0.11 (-0.03, 0.25) 0.01 (-0.19, 0.18)
Brampton/Mississauga
Adjusted
Unadjusted model ([double
models ([dagger]) dagger)]
[beta] 95% CI [beta] 95% CI
Material
deprivation
Least deprived -- --
Middle -2.46 (-10.54, 5.62) -0.86 (-9.06, 7.34)
Most deprived -2.90 (-9.64, 3.85) 0.52 (-7.14, 8.17)
Urban form
Population
density
(1000 per
[km.sup.2]) -0.98 * (-1.91, -0.05) -0.66 (-1.70, 0.38)
% Detached
housing 0.13 * (0.01, 0.25) 0.06 (-0.11, 0.23)
% Public
transit -0.53 (-1.08, 0.03) -0.26 (-0.95, 0.42)
Hamilton
Adjusted
Unadjusted model ([double
models ([dagger]) dagger)]
[beta] 95% CI [beta] 95% CI
Material
deprivation
Least deprived -- --
Middle 8.01 (-2.12, 18.13) 6.44 (-3.96, 16.84)
Most deprived 6.76 (-1.59, 15.10) 5.25 (-4.78, 15.28)
Urban form
Population
density
(1000 per
[km.sup.2]) -0.35 (-1.98, 1.28) -0.05 (-2.05, 1.96)
% Detached
housing 0.11 (-0.03, 0.26) 0.27 ** (0.08, 0.46)
% Public
transit 0.48 (-0.16, 1.12) 0.94 * (0.01, 1.86)
users
[beta]=parameter estimate (unstandardized slope coefficient); 95%
CI=95% confidence interval.
([dagger]) Estimates obtained from separate unadjusted Tobit
regression models.
([double dagger]) All estimates mutually adjusted for all other
variables listed.
* p<0.05, ** p<0.01.
